Comparison of cluster software


The following tables compare general and technical information for notable computer cluster software. This software can be grossly separated in four categories: Job scheduler, nodes management, nodes installation and integrated stack.

General information

SoftwareMaintainerCategoryDevelopment statusLatest releaseArchitectureOCSHigh-Performance / High-Throughput ComputingLicensePlatforms supportedCost
Amoeba active development
Base One Foundation Component Library
DIETINRIA, SysFera, Open SourceAll in oneGridRPC, SPMD, Hierarchical and distributed architecture, CORBAHTC/HPCUnix-like, Mac OS X, AIX
Nodes management developedv23.0Windows 2012R2/2016/2019/2022 and 8+, RHEL 7/8/9, CentOS 7, Ubuntu 16.04/18.04/20.04/22.04, SUSE [Linux Enterprise Server|SLES] 15.4Cost
Enduro/XMavimax, Ltd.Job/Data Scheduler developedSOA GridHTC/HPC/HAGPLv2 or CommercialLinux, FreeBSD, MacOS, Solaris, AIXFree / Cost
GangliaMonitoring developed Unix, Linux, Microsoft Windows NT/XP/2000/2003/2008, FreeBSD, NetBSD, OpenBSD, DragonflyBSD, Mac OS X, Solaris, AIX, IRIX, Tru64, HPUX.
Grid MPUniva Job Scheduler active developmentDistributed master/workerHTC/HPCWindows, Linux, Mac OS X, Solaris
Apache MesosApache developedLinux
Moab Cluster SuiteAdaptive ComputingJob Scheduler developedHPCLinux, Mac OS X, Windows, AIX, OSF/Tru-64, Solaris, HP-UX, IRIX, FreeBSD & other UNIX platforms
NetworkComputerRuntime Design Automation developedHTC/HPCUnix-like, Windows
OpenClusterSchedulerOpen Cluster Schedulerall in one developed9.0.8 HTC/HPCSISSL / Apache LicenseLinux, FreeBSD, Solaris
OpenHPCOpenHPC projectall in one developedv2.61 HPCLinux
OpenLava Formerly TeraprocJob SchedulerHalted by injunctionMaster/Worker, multiple admin/submit nodesHTC/HPCIllegal due to being a pirated version of IBM Spectrum LSFLinux
PBS ProAltairJob Scheduler developedMaster/worker distributed with fail-overHPC/HTCAGPL or ProprietaryLinux, Windows or Cost
Proxmox Virtual EnvironmentProxmox Server SolutionsComplete developedLinux, Windows, other operating systems are known to work and are community supported
Rocks Cluster DistributionOpen Source/NSF grantAll in one developed HTC/HPCCentOS
Popular Power
ProActiveINRIA, ActiveEon, Open SourceAll in one developedMaster/Worker, SPMD, Distributed Component Model, SkeletonsHTC/HPCUnix-like, Windows, Mac OS X
RPyCTomer Filiba developed*nix/Windows
SLURMSchedMDJob Scheduler developedv23.11.3 HPC/HTCLinux/*nix
Spectrum LSFIBMJob Scheduler developedMaster node with failover/exec clients, multiple admin/submit nodes, Suite addOnsHPC/HTCUnix, Linux, Windows and Academic - model - Academic, Express, Standard, Advanced and Suites
Oracle Grid Engine (Sun Grid Engine, SGE)AltairJob Scheduleractive Development moved to Altair Grid EngineMaster node/exec clients, multiple admin/submit nodesHPC/HTC*nix/Windows
Some Grid Engine / Son of Grid Engine / Sun Grid EnginedaimhJob Scheduler developed Master node/exec clients, multiple admin/submit nodesHPC/HTC*nix
SynfiniWayFujitsu developedHPC/HTCUnix, Linux, Windows
Techila Distributed Computing EngineAll in one developedMaster/worker distributedHTCLinux, Windows
TORQUE Resource ManagerAdaptive ComputingJob Scheduler developedLinux, *nix
TrinityXAll in one developedv15 HPC/HTC v3Linux/*nix
UniClusterUnivaAll in OneFunctionality and development moved to UniCloud
UNICORE
XgridApple Computer
WarewulfProvision and clusters management developedv4.6.4 HPCLinux
xCATProvision and clusters management developedv2.17.0 HPCEclipse Public LicenseLinux
SoftwareMaintainerCategoryDevelopment statusLatest releaseArchitectureHigh-Performance/ High-Throughput ComputingLicensePlatforms supportedCost

Table explanationSoftware: The name of the application that is described

Technical information

SoftwareImplementation LanguageAuthenticationEncryptionIntegrityGlobal File SystemGlobal File System + KerberosHeterogeneous/ Homogeneous exec nodeJobs priorityGroup priorityQueue typeSMP awareMax exec nodeMax job submittedCPU scavengingParallel jobJob checkpointing
Enduro/XC/C++OS AuthenticationGPG, AES-128, SHA1 cluster Posix FS cluster Posix FS HeterogeneousOS Nice levelOS Nice levelSOA Queues, FIFOOS LimitsOS Limits
HTCondorC++GSI, SSL, Kerberos, Password, File System, Remote File System, Windows, Claim To Be, AnonymousNone, Triple DES, BLOWFISHNone, MD5None, NFS, AFSHeterogeneousFair-share with some programmabilitybasic tested ~10000?tested ~100000?MPI, OpenMP, PVM
PBS ProC/PythonOS Authentication, Munge, e.g., NFS, Lustre, GPFS, AFSLimited availabilityHeterogeneousFully configurabletested ~50,000MillionsMPI, OpenMP
OpenLavaC/C++OS authenticationNFSHeterogeneous LinuxConfigurable, supports preemption based on priority
SlurmCMunge, None, KerberosHeterogeneousMultifactor Fair-sharetested 120ktested 100k
Spectrum LSFC/C++Multiple - OS Authentication/Kerberos - GPFS/Spectrum Scale, NFS, SMB - GPFS/Spectrum Scale, NFS, SMBHeterogeneous - HW and OS agnostic Policy based - no queue to computenode bindingPolicy based - no queue to computegroup bindingBatch, interactive, checkpointing, parallel and combinations and GPU aware > 9.000 compute hots> 4 mio jobs a day, supports preemption based on priority, supports checkpointing/resume, fx parallel submissions for job collaboration over fx MPI, with support for user, kernel or library level checkpointing environments
TorqueCSSH, mungeNone, anyHeterogeneousProgrammabletestedtested
SoftwareImplementation LanguageAuthenticationEncryptionIntegrityGlobal File SystemGlobal File System + KerberosHeterogeneous/ Homogeneous exec nodeJobs priorityGroup priorityQueue typeSMP awareMax exec nodeMax job submittedCPU scavengingParallel jobJob checkpointing

Table ExplanationSoftware: The name of the application that is describedSMP aware:
  • * basic: hard split into multiple virtual host
  • * basic+: hard split into multiple virtual host with some minimal/incomplete communication between virtual host on the same computer
  • * dynamic: split the resource of the computer on demand